Dates, Calendars and Conversion

A date is a number, and a number carries no information about which system produced it. Foreign documents routinely present dates in a calendar the reader does not use, in a field order that reverses between conventions, or as a figure that was never known precisely in the first place.

Calendars in official use

The Gregorian calendar is not universal. Several countries maintain civil registration in a different system, and documents issued there carry that system's dates without any indication that conversion is required.

Lunar calendars drift against solar ones. A lunar year is shorter, so the relationship between a lunar date and a Gregorian one shifts every year and cannot be handled with a fixed offset.

Dual dating is common. Many documents show both systems, which removes the problem entirely, and the presence of two date fields on a certificate is worth looking for before assuming anything.

Era systems restart. Calendars counting from the accession of a monarch or the founding of a state restart their numbering, so a two-digit year is meaningless without the era.

Conversion is arithmetic, not judgment. Published tables and official converters exist, and a conversion supported by a citation to one of them is far stronger than a figure asserted in a statement.

Field order, and where it bites

Two conventions, both common. Day-month-year and month-day-year are each used by large parts of the world, and a numeric date is ambiguous whenever both components are twelve or less.

Above twelve it resolves itself. A date containing a number above twelve in either position fixes the order, which means most dates in a document establish the convention used throughout it.

Look at the whole document. One unambiguous date on a certificate settles the format for every other date on it, and that inference is usually available.

Systems reverse silently. Data entered in one convention and displayed in another produces a wrong date with no warning at all, which is a frequent source of apparent discrepancy.

Written months remove it. A date rendered with the month spelled out cannot be misread, which is why competent translations write dates out rather than reproducing the numerals.

Dates that were never known precisely

Estimated at registration. Where a birth was registered years afterward, the date recorded is what an informant believed, and in some systems the practice of assigning a default day is explicit.

The first of the month, or the first of the year. A striking concentration of first-of-January birth dates in some registries reflects an administrative default rather than a remarkable coincidence.

Age recorded instead of a date. Some older records state an age rather than a date of birth, from which a year has been calculated afterward, with an inherent margin of a year either way.

Documents disagree accordingly. Two records estimating the same unknown date will differ, and that difference is not evidence that either is false, as documents that contradict the record sets out.

Acknowledge rather than resolve. A statement explaining that the date was estimated at registration is more credible than an argument that one of the two documents must be correct.

Evidencing a conversion

Show the working. The original date exactly as written, the calendar it belongs to, the converted date, and the source relied on for the conversion. Four lines, and they answer the question before anybody asks it.

Prefer an official converter. Several governments publish conversion tools for their own calendars, and a citation to the issuing country's own source is difficult to argue with.

Keep the original visible. A filing that silently substitutes the converted date creates a document disagreeing with its source, which is the problem described in names spelled differently abroad.

Have the translator handle it. A competent translation renders the date in its original form and gives the Gregorian equivalent in brackets, which puts both versions on the same page and removes any later argument about who converted what.

Watch for double conversion. A date converted once by a registry and again by a translator produces an error that is hard to spot and easy to make.

Working with the dates

Establish the convention first. Before reading any date on a foreign document at all, establish which calendar and which field order the issuing system used at that time. Everything afterward depends on that answer.

Write dates out in every filing. Month spelled in full, four-digit year, which removes the ordering ambiguity permanently at no cost.

Reconcile the whole file at once. Listing every date across every document, in both forms, catches the conversions that were done inconsistently.

Expect matching to fail. Institutions search on exact dates, so a converted date entered against an unconverted record returns nothing rather than a near match.

Explain rather than correct. Amending a foreign register over a date is as slow as amending one over a name, and receiving bodies accept explanation, as civil registries that work differently describes.

Dates carry an unearned appearance of precision. A number on a certificate looks like a fact, and it may be a conversion, a reversal, or a figure somebody supplied because the form required one.

The two mechanical problems, calendar and ordering, are entirely solvable. Conversion is arithmetic that can be cited, and ordering resolves itself from any unambiguous date elsewhere on the same document.

The third problem is not solvable and should not be presented as though it were. Where a date was estimated at registration, the honest course is to say so and to explain the practice, which decision-makers accept readily.

Every filing prepared from foreign documents benefits from one page listing every date in both forms. It takes half an hour, it catches inconsistent conversions, and it becomes the explanation if anybody asks.

Finally, the same mechanical caution applies to searches. A converted date submitted against an unconverted register produces a nil return, and the response to a nil return should always include the alternative form.